LIPIcs.ECOOP.2019.14.pdf
- Filesize: 0.73 MB
- 25 pages
Document
NumLin: Linear Types for Linear Algebra
Authors
Dhruv C. Makwana 
,
Neelakantan R. Krishnaswami
-
Part of:
Volume:
33rd European Conference on Object-Oriented Programming (ECOOP 2019)
Series: Leibniz International Proceedings in Informatics (LIPIcs)
Conference: European Conference on Object-Oriented Programming (ECOOP) - License:
Creative Commons Attribution 3.0 Unported license
- Publication date: 2019-07-10
File
Document Identifiers
Author Details
Acknowledgements
We would like to thank Stephen Dolan for his advice and support with the implementation and evaluation of NumLin. We would also like to thank the (anonymous) reviewers for their feedback and suggestions.
Cite AsGet BibTex
Dhruv C. Makwana and Neelakantan R. Krishnaswami. NumLin: Linear Types for Linear Algebra. In 33rd European Conference on Object-Oriented Programming (ECOOP 2019). Leibniz International Proceedings in Informatics (LIPIcs), Volume 134, pp. 14:1-14:25, Schloss Dagstuhl - Leibniz-Zentrum für Informatik (2019)
https://doi.org/10.4230/LIPIcs.ECOOP.2019.14
https://doi.org/10.4230/LIPIcs.ECOOP.2019.14
Abstract
We present NumLin, a functional programming language whose type system is designed to enforce the safe usage of the APIs of low-level linear algebra libraries (such as BLAS/LAPACK). We do so through a brief description of its key features and several illustrative examples. We show that NumLin’s type system is sound and that its implementation improves upon naïve implementations of linear algebra programs, almost towards C-levels of performance. By doing so, we demonstrate (a) that linear types are well-suited to expressing the APIs of low-level linear algebra libraries accurately and concisely and (b) that, despite the complexity of prior work on it, fractional permissions can actually be implemented using simple, well-known techniques and be used practically in real programs.
Subject Classification
ACM Subject Classification
- Theory of computation → Program specifications
Keywords
- numerical
- linear
- algebra
- types
- permissions
- OCaml
Metrics
- Access Statistics
-
Total Accesses (updated on a weekly basis)
0PDF Downloads